[20a-C101-1~13] 7.3 Micro/Nano patterning and fabrication

[20a-C101-5] Pattern prediction in EB lithography using molecular simulation and deep learning

〇Bunta Inoue¹, Yoshihiko Hirai², Masaaki Yasuda² (1.Osaka Pref. Univ., 2.Osaka Met. Univ.)

Keywords:Electron beam lithography, Deep learning, Molecular simulation

We constructed a neural network that describes the correlation between material and process conditions and pattern shape in electron beam lithography using the results of molecular simulation as training data. Patterns were predicted under conditions not included in the training data using the constructed system. The predicted shape was able to reproduce the tendency obtained by the molecular simulation well. In addition, the constructed system was able to speed up shape prediction by 150 times compared to analysis by molecular simulation.